Department of Industrial Engineering & decision Analytics [Joint IEDA/ISOM] seminar  - Managing User Engagement under Reference Effects

Effectively managing user engagement is a central challenge for online platforms. While this issue shares similarities with traditional operations management problems, it also presents unique complexities. In this talk, I will present two projects that address user engagement in video games using dynamic programming, focusing on how users' behaviors are influenced by reference effects.

     In the first project, we investigate highly addictive video games, aiming to address digital addiction while maintaining profitability. By modeling users’ habit-forming behaviors, we characterize and compare optimal fixed and dynamic pricing strategies across various scenarios. Our results show that dynamic pricing not only increases profits but also helps mitigate long-term addiction, thus benefiting both firms and users. Furthermore, we analyze three addiction control policies—consumption limits, price floors, and consumption taxes—and evaluate their efficiency in balancing profit and addiction control. We find that consumption limits or taxes are most effective, with experience variation playing a key moderating role.

     The second project focuses on video games with extremely high churn rates. Here, our objective is to enhance player retention and lifetime value by dynamically adjusting game difficulty. We develop a practical, data-driven methodology based on the reference effect model and conduct A/B tests in two mobile games. Compared to human-designed difficulty settings, our policy increases player retention by 203% and 297%, and boosts lifetime revenue by 23% and 7%, respectively.